Comment by HPsquared

5 days ago

While we're on noise-cancelling headphones, this reminds me of a recent experience on a train (in the quiet carriage, of course) wearing my noise-cancelling headphones. There was a toddler crying and screaming loudly and the sound seemed to cut right through the headphones. I wonder if they're somehow tuned to not block certain sounds (Sony WH-1000XM4).

I think it's more that noise cancellation is more difficult at higher frequencies. You have to match the waveforms much more precisely to get effective cancellation at high frequencies, so active cancellation is mostly going to work in the low range. Works great for a lot of steady machinery sounds like the rumble of a train or airplane, but does little for high-pitched voices.

When I wear Bose noise-cancelling headphones on airplanes I can hear the female flight attendants much better than when I'm not wearing headphones. The noise-cancelling filters out lower frequencies and constant hum (wind, engines), but not the higher-pitched frequencies (which is why the female voices are easier to hear, and, as you mentioned, crying toddlers). So, when the announcement asks people to remove headphones while they explain the security setup it actually has the opposite effect for me - it makes it more difficult to hear what they're saying if I remove my headphones.

Note that there are different kinds of noise-cancelling headphones. The ones pilots wear are different, and then you have those made for hunters - they are focused at reacting very fast to sounds from shooting.

  • My Sony ones have an actual setting for this, the ambient sound mode has a "focus on voices" option and yes, it does help me pick up voices more clearly than usual if there's background noise.

They're not, it's just the physics of sound cancelation. Some frequencies, especially higher, irregular ones are harder to block, just because the microprocessor doesn't have enough time to react

noise cancelling works best for constant sounds (like the thrum of an engine or a murmur of the crowd) so the anc processor can "compensate" for that sound. If a toddler was crying irregularly, with "wahs" at longer bursts, it's possible that the ANC didn't have time for it's time window to compensate for the sound. Aka the sound started and stopped before a particular time window.

not sure if it is by design, but a larger window would potentially decrease computational load since it doesn't need to recalibrate and reprocess at a higher rate.

  • Probably white noise would cover those but some people can’t take white noise. I actually love white noise, it calms me and helps me stay focused on the task at hand

I disabled noise-cancelling on my headphones because I actually found it make the sound of my kid crying very uncomfortable.

I honestly think that noise-cancelling is a great idea for a technology, and is basically required from a marketing perspective, but not all that helpful in practice. Sound isolation isn't is sexy, but it works much better.

I would think there'd be a lot of utility in not blocking smoke alarm sounds, for example. The screeching from distressed infants can similarly be both very loud and very high pitched.

  • Infants scream at the resonant frequency of the human ear canal, about 3000 hz

    • I'm trying to search this and it seems the numbers for both (ear canal and infants cries) are all over the place. Do you have a specific article about the connection?

      Interesting if true...

      1 reply →